Course › Module 12 · Rollout and capstone

Your AI recruiting toolkit and next steps

Module 12, Lesson 5  ·  4 min read ·  Updated 21 September 2026

Module 12 · Lesson 5

You now have a working method rather than a collection of tools. Here is what you built, what this course deliberately left out, and where to go next.

What you have built

FromYou have
Module 2An intake method and a job blueprint that everything else runs on
Module 3Searches you can repeat, and a reusable prompt library
Module 4Screening rules, a way to check how well CVs are read, and the disagreement check
Module 5A match profile, and a test for whether an explanation is real
Module 6A five-message sequence and a way to check it
Module 7A structured interview loop and a debrief that is not just group opinion
Module 8A hiring process configured so the numbers mean something
Module 9A workflow with one real human check, tested against hostile input
Module 10A weekly dashboard and your own conversion rates
Module 11A risk register, candidate notices, and records you could defend

None of that depends on any particular product. It is a way of working that happens to use software.

What this course left out

  • Executive search. Different work, driven by relationships and market knowledge rather than volume.
  • High-volume hourly hiring. Thousands of applications for identical jobs is a different problem, closer to operations than to what is here.
  • Internal mobility. Similar principles, different politics, different data.
  • Assessment design. Building valid tests is a specialist field. Use well-established tests rather than inventing your own.
  • One country's employment law in depth. Module 11 is a map, not legal advice.

What to do next

  1. Run the capstone on a second job. The first run teaches you the method. The second teaches you which parts of it you actually kept.
  2. Replace the planning numbers with your own. Every figure in Module 3 was a starting point. Yours are better.
  3. Put the reviews in a calendar. The risk register and bias check every three months, on a date.
  4. Teach one part to someone else. The scorecard method is the best place to start, and teaching it is how you find out what you actually understood.

The honest limit of doing this by hand

Everything here works with a spreadsheet and careful attention. That is genuinely true for one or two jobs at a time.

It stops being true at volume. When you are running fifteen jobs, the parts that get dropped first are exactly the parts that make the process defensible: the disagreement check, saving your criteria with each shortlist, rejections with reasons, the bias check. Not because anyone decided to stop caring — because they are the steps with no immediate consequence for skipping them.

That is the real argument for tooling. Not that software thinks better than you. That it makes the careful version survive a real workload.

If you want to run this at volume, we build HireGen, the tool this course refers to when it describes an interface. It does the parts this course spent most of its time on: reading CVs with the fields visible, screening against rules you write, ranking with quoted evidence, and the records Module 11 asks for. The free plan covers everything in these exercises.

See what HireGen does

It is not the only tool that will do this. The test from Module 1, Lesson 4 still applies: can you see the extracted fields, can you see why a candidate ranked where they did, can you export the records. Ask any supplier, including us.

One last thing

The single habit worth keeping from all of this: after every screening run, open the three highest-scoring people who were rejected.

It takes five minutes. It is what separates screening from handing the job over. And it is the first thing everyone stops doing.

Thank you for reading

If something here was wrong, unclear, or did not survive contact with your actual hiring process, we would like to know — that is how the next version gets better.

Common questions

How do you roll out AI recruiting without breaking things?

Spend the first 30 days measuring and fixing data, deploying nothing — you need a before picture and it cannot be reconstructed later. Days 31 to 60, do one role family completely. Days 61 to 90, connect the handoffs and extend. Leave automated rejection, candidate-facing messaging and learned ranking alone in month one.

How do you get hiring managers to use AI screening?

Make the scorecard theirs. If they wrote the criteria in the intake meeting, the screening applies their judgement at scale rather than replacing their taste — that framing does more than everything else combined. Then show them the rejects rather than the shortlist; people trust a system more once they have seen its mistakes caught.

How do you build a business case for AI recruiting tools?

Attach every saving to a decision. 'Twelve hours a week of productivity' gets discounted to zero; 'nine hours a month, which lets one recruiter carry the two extra requisitions planned for Q3 without a hire' does not. Agency fees avoided is the strongest line because the counterfactual is a real invoice. Show the downside case before you are asked for it.

Check yourself

Four questions. Nothing is recorded anywhere but your own browser — this is for you, not for a score.

  1. 1What should you do in the first 30 days of a rollout?

  2. 2What most reliably gets a hiring manager to adopt AI screening?

  3. 3Which time saving will a CFO actually fund?

  4. 4Which habit is the first to be dropped as volume rises, and why does that matter?